Who it’s for: Businesses with 20–200 staff that produce regular reports for owners, boards or clients
Report and summary drafting from your own data
Drafts weekly and monthly reports from your own systems, with every line linked to its source and anything unsupported flagged for review.
The problem
Every week or month someone pulls numbers from a spreadsheet, the CRM and the accounting system, then turns them into a report. It takes hours, it is tedious, and it tends to be written by whoever is least busy. A generic AI summary is no better, because nobody can see where a statement came from or whether it was made up.
What we build
- A workflow that gathers data from your spreadsheets, CRM, accounting exports and logs on a schedule
- A draft report or summary in your format, written in plain language
- Source links on every line, with unsupported lines flagged for review and never filled in
How it works
Your report becomes the template
We take your last few reports and agree the structure, the measures and the wording with the person who writes them. We also agree which systems count as the source for each measure.
Data is gathered
On a schedule, from exports or direct connections. We confirm the options for your systems during discovery, before we quote.
A draft with sources
The system writes the draft from the data, linking each line to the source that supports it. If it cannot find a source for a line, it says so and flags the line for review. It does not fill the gap.
A person reviews
The owner of the report checks the flagged lines, edits the wording and approves it. Only then is it shared.
What it looks like
Illustrative example of a weekly operations summary for a fictional business. Three lines link to a sales export, a dispatch log and a CRM report. One line about a stock count variance has no source and is flagged for review.
What you receive
- The working reporting workflow, connected to your data sources
- Your report format and measures, which you own and can change
- A test set of past reporting periods, with results against it
- A one-page system description: what it reads, what it flags, where a person reviews and where it runs
- Training for the people who will use it, and a runbook
How we check accuracy
- A test set of your past reports, with the source data for each period
- A pass mark agreed with the report owner before go-live
- A shadow run alongside your usual reporting before anyone relies on it
- Regular quality checks against the same test set, reported to you
What changes
- A consistent draft ready when the reporting day arrives
- Reviewers check lines against sources instead of writing from scratch
- A clear record of where each statement came from
What stays with you
What the report says, who sees it and any decision made from it.
In practice
We build report drafting around your data and your format. The system is designed to save the gathering and first-draft work, and to show its sources so a reviewer can check each line quickly.
It doesn’t replace the judgement of the person who owns the report. It gives them a checked starting point.
Each build is fitted to your systems and the way your readers like to receive information.
Questions
What if a number has no source?
The line is flagged as needing review. The system is built not to fill gaps, so you see what it could not support.
Which reports work best?
Regular reports built from the same systems each time, such as weekly operations, monthly management and project status updates. We confirm the fit during discovery.
Can it match our format?
Yes. We start from your recent reports and keep the structure and wording your readers expect.
Does it send the report?
No. It prepares a draft. A person reviews it and decides who receives it.
Is this part of the retainer?
Yes. It’s one of the processes we can take on in a month, or a fixed-scope build if you prefer.
Where is our data processed?
For client engagements we build on AWS in Australia (Sydney region where available); any exception only with your written agreement.
Related
AI document processing
Reads supplier invoices, purchase orders, forms and contracts, checks them against your rules and flags exceptions for a person to decide.
Read, check and flag documents
Knowledge assistant
Answers staff questions from your own policies, procedures and product information, citing the document and section, or saying when it can’t find one.
Answers from your own documents